Table of Contents

Marketing analysis of the Procrastinate Pro+ application

Research purpose

To figure out the reason of the Procrastinate Pro+ entertainment application profits decrease in the last few months. And help the company to bring back the profit growth.

Research description

Despite huge investments in advertising, the company has been suffering losses for the past few months. The task is to understand the reasons of profit decrease and help the company to growt the profit.
Data contains users registered from May 1 to October 27, 2019:
server log with data about their visits,
logs of their purchases for this period,
advertising expenses.
There's a heed to find out: where users come from and what devices they use,
how much does user acquisition costs from various advertising channels;
how much money does each client bring,
when the cost of a client acquisition pays off,
what factors prevent customers acquisition.

Data preprocessing

visits_info_short.csv file stores the server log with information about site visits:

orders_info_short.csv — information about purchases

costs_info_short.csv — information about advertising expenses

Data preprocessing

Dataset Costs

The data is clean, without omissions and duplicates.
The dt column will be converted to the datetime format.
Let's bring the columns to the lower case.

Dataset Orders

The data is clean, without omissions and duplicates.
Let's bring the columns to the lower case.
The Event Dt column will be converted to the datetime format.

Dataset Visits

The data is clean, without omissions and duplicates.
Let's bring the columns to the lower case.
The Event Dt column will be converted to the datetime format.


The results of the preview of the Costs, Orders, Visits datasets:
Omissions and duplicates were not found in any dataset.
Column names are reduced to lowercase.
Columns with dates are converted to datetime format.

Functions for calculating and analyzing LTV, ROI, retention and conversion

Function for creating user profiles get_profiles

Function for calculating retention get_retention

Conversion calculation function get_conversion

Function for calculation LTV и ROI get_ltv

Function for smoothing the frame filter_data

Function for visualizing retention plot_retention

Conversion visualization function plot_conversion

Visualization function LTV и ROI plot_ltv_roi

Data analysis

Creating user profiles

Determining the minimum and maximum dates for users acquisition

Distribution of paying users by country

Distribution of paying users by device

Distribution of paying users by advertising channels


Advertising costs in the US are 6 times more than 90 thousand units than in France, Germany and the UK combined.
Advertising costs in France, Germany and the UK are about the same - 5 thousand cu.
At the same time, the number of US users is only 2 times more than the alliance of France, Germany and UK.
The US also has the highest percentage of paying users - 70%, compared to 40% in other countries.


iPhone users cost more (44,000 USD) than Mac and Android (23,000 USD) and PC (15,000 USD). Paying users are distributed fairly evenly across the platforms - 60-50%.
Mac conversion leader is 6%, PC users are the least converted - 5%


Spending on TipTop and FaceBoom is 4 times more than on all other channels.
The number of users asquired through TipTop and FaceBoom is greater than the number of users of all other channels.
Notable leaders of the conversion in paying users - FaceBoom, AdNonSense, lambdaMediaAds, TipTop Interestingly, FaceBoom's conversion rate is 12%, which is more than TipTop's - 9.6%

Marketing

Distribution of expenses by sources


The cost of acquisition one user from each source


Evaluation of the payback of advertising

The total payback of advertising. LTV and ROI charts and LTV, CAC and ROI dynamics


The total payback of advertising through advertising channels. LTV and ROI charts and LTV, CAC and ROI dynamics


The total payback of advertising by country. LTV and ROI charts and LTV, CAC and ROI dynamics


Payback of advertising in the USA through advertising channels


Payback of advertising in the UK, Germany, France through advertising channels

Conversion and retention charts by device, country, and advertising channels







Facebook channel conversion and retention charts by device


AdNonSense channel conversion and retention charts by device


TipTop channel conversion and retention charts by device


Project summary


It seems that FaceBoom is the worst, but at the same time quite expensive advertising channel. It should be abandoned.
And TipTop had an abnormal price increase - a multiple jump every month, perhaps we hit the peak of the channel's popularity. At the same time, TipTop supplies good users, but due to the increase in cost, the costs for them do not pay off.
It is necessary to discuss cost reduction, it is possible to increase the volume of purchases at the expense of part of the budget on FaceBoom.

Which tells us that the US has very responsive users, but there is a failure in retention, which results in a sharp increase in CAC. It is worth checking the technical side of retention for errors at all stages of the "user's life"

Recommendations


Based on the charts of marketing metrics 8.3.1-2, we can recommend more promising channels than the problematic FaceBoom and expensive TipTop:

United States:

UK, Germany, France